Journal of ELECTRONIC MATERIALS

DOI: 10.1007/s11664-017-5940-8
 2017 The Minerals, Metals & Materials Society

Effects of Lu and Tm Doping on Thermoelectric Properties

of Bi2Te3 Compound

MAXIM YAPRINTSEV,1 ROMAN LYUBUSHKIN,1 OXANA SOKLAKOVA,1

and OLEG IVANOV 1,2,3

1.—Belgorod State University, Belgorod, Russian Federation 394015. 2.—Voronezh State Tech-
nical University, Voronezh, Russian Federation 394026. 3.—e-mail: Ivanov.Oleg@bsu.edu.ru

The Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 thermoelectrics of n-type con-

ductivity have been prepared by the microwave-solvothermal method and
spark plasma sintering. These compounds behave as degenerate semicon-
ductors from room temperature up to temperature Td  470 K. Within this
temperature range the temperature behavior of the specific electrical resis-
tivity is due to the temperature changes of electron mobility determined by
acoustic and optical phonon scattering. Above Td, an onset of intrinsic con-
ductivity takes place when electrons and holes are present. At the Lu and Tm
doping, the Seebeck coefficient increases, while the specific electrical resis-
tivity and total thermal conductivity decrease within the temperature 290–
630 K range. The increase of the electrical resistivity is related to the increase
of electron concentration since the Tm and Lu atoms are donor centres in the
Bi2Te3 lattice. The increase of the density-of-state effective mass for conduc-
tion band can be responsible for the increase of the Seebeck coefficient. The
decrease of the total thermal conductivity in doped Bi2Te3 is attributed to
point defects like the antisite defects and Lu or Tm atoms substituting for the
Bi sites. In addition, reducing the electron thermal conductivity due to form-
ing a narrow impurity (Lu or Tm) band having high and sharp density-of-
states near the Fermi level can effectively decrease the total thermal con-
ductivity. The thermoelectric figure-of-merit is enhanced from  0.4 for un-
doped Bi2Te3 up to  0.7 for Bi1.9Tm0.1Te3 and  0.9 for Bi1.9Lu0.1Te3.

Key words: Bi2Te3 compound, rare earth element doping, thermoelectric

properties

INTRODUCTION thermal conductivity with contributions from crys-

tal lattice and carriers. Therefore, lower k and q,
Currently, bismuth telluride, Bi2Te3 and Bi2Te3-
and higher S, should be combined for material to
based compounds are the best thermoelectric mate-
reach a higher ZT value. A number of investigations
rials for applications around room temperature.1–3
using various physical and technological approaches
Unfortunately, thermoelectric efficiency of these
have been carried out to improve the thermoelectric
compounds defined by the thermoelectric figure-of-
properties of Bi2Te3.5–13 A doping is one of obvious
merit, ZT, remained too low (ZT £ 1) until now. As
and promising ways to optimally combine the S, q
is known,4 the dimensionless figure-of-merit is
and k values and enhance ZT of materials.14–16
expressed as (S2/qk)T, where T is the absolute
Recently, it was found that rare earth element
temperature, S is the Seebeck coefficient, q is the
(Lu, Ce, Sm, Er, La, etc.) doping can be used to
specific electrical resistivity, and k is the total
enhance the thermoelectric performance of
Bi2Te3.17–24 There are several effects of the rare
earth element doping: (1) the increase of carrier
(Received August 21, 2017; accepted November 7, 2017) concentration due to donor-like effects at rare earth
 Yaprintsev, Lyubushkin, Soklakova, and Ivanov

elements substituting for Bi site in the Bi2Te3 The specific electrical resistivity, q, and the
lattice; (2) the increase of the electron and phonon Seebeck coefficient, S, were measured by using a
scattering by point defects forming in the crystal ZEM-3 system.
lattice at the doping; (3) the increase of the Seebeck To determine the type, concentration, n, and
coefficient by enhanced electron states forming near mobility, lH, of the majority charge carriers, the
the Fermi level; and (4) the additional increase of Hall effect was studied by a Cryogenic Free system.
carriers scattering by localized magnetic moments A TC-1200 system was applied to determine the
of some rare earth elements. thermal conductivity, k, by the laser flash method.
The aim of this paper is to study effects of rare Measurement by the laser flash method finds ther-
earth element doping on the thermoelectric proper- mal diffusivity based on the temperature change
ties of the Bi2Te3 compound. Lutetium, Lu, and with time of the back side of a sample after heating
tullius, Tm, were used as dopants to prepare the the front side with laser beam instantaneously. The
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 compositions. thermal conductivity of a sample is found from the
thermal diffusivity a, the specific heat Cp and the
EXPERIMENTAL density d using the k = aÆCpÆd equation. For the
measurement of the specific heat, a standard sam-
Microwave-solvothermal synthesis was applied to
ple is used at room temperature to find the absorbed
prepare the starting powders of the Bi2Te3,
energy of the sample. Then the specific heat of an
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 compositions. As is
unknown sample is found at a target temperature
known, compared with the conventional methods,
by comparing the temperature changes of the
the microwave-assisted heating technique has the
sample at room temperature and target tempera-
advantages of very short time of synthesis, simplic-
ture, assuming that the same heat capacity is
ity and energy efficiency, small particle size of the
absorbed by the unknown sample.
products and narrow particle size distribution, and
high purity of powder.25,26
RESULTS AND DISCUSSION
Analytically pure chemicals were used for the
synthesis (bismuth oxide, Bi2O3, tellurium oxide, The XRD patterns for the bulk Bi2Te3,
TeO2, lutetium oxide, Lu2O3, tullius oxide, Tm2O3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 compounds taken
ethylene glycol, nitric acid and N,N- at room temperature are shown in Fig. 1. According
dimethylformamide). to the XRD analysis, all sintered materials are of
The Bi2O3, TeO2 and Lu2O3 or Tm2O3 oxides the single hexagonal phase characteristic for pure
taken in a stoichiometric ratio were dissolving in a Bi2Te3. The Lu and Tm atoms are believed to
mixture of concentrated nitric acid and ethylene incorporate to the Bi2Te3 lattice changing the lattice
glycol. Then, N,N-dimethylformamide was added in parameters. The lattice a and c parameters calcu-
the mixture after dissolving. The microwave-as- lated by the Rietveld refinement are listed in
sisted reaction was carried out in a MARS-6 Table I. In fact, the a and c changes are very small
microwave reactor with a power of 1000 W and at and close to accuracy of the XRD analysis. So, just
2.45 MHz working frequency. The synthesis was the weak effect of the Lu and Tm doping on the
carried out for 15 min at temperature of 463 K and Bi2Te3 structure could be found in XRD phases.
pressure of 40 bars.
The spark plasma sintering method, by using a
SPS-25/10 system, was applied to sinter the bulk
materials at a pressure of 40 MPa, a temperature of
683 K and a sintering time of 5 min.
The densities of the bulk samples were measured
by the Archimedes’ method.
X-ray diffraction (XRD) analysis was performed
by a Rigaku Ultima IV diffractometer with CuKa-
radiation to determine the phase composition of the
starting powders and the bulk materials.
To determine the correct elemental composition of
materials prepared, a Shimadzu ICP (Inductively
Coupled Plasma) emission spectrometer ICPE-9000
was applied.
A scanning electron microscope (SEM), a Nova
NanoSEM 450, was used to study the microstruc-
ture features of the bulk samples. The energy
dispersive x-ray spectroscopy (EBSD) method was
applied to map the distribution of chemical elements
for the bulk Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3
compounds. Fig. 1. XRD patterns of Bi2Te3 (a), Bi1.9Lu0.1Te3 (b) and
Bi1.9Tm0.1Te3 (c).
Effects of Lu and Tm Doping on Thermoelectric Properties of Bi2Te3 Compound

Table I. Lattice parameters and elemental compositions of Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3
Compound a (Å) c (Å) Bi (at.%) Te (at.%) Lu (at.%) Tm (at.%)

Bi2Te3 4.385 30.476 40.00 60.00 – –

Bi1.9Tm0.1Te3 4.387 30.484 38.04 60.00 – 1.96
Bi1.9Lu0.1Te3 4.388 30.481 38.03 60.00 1.98 –

Weakness of this effect should be attributed to a

small difference between ionic radii of Lu (1.001 Å),
tullius (1.020 Å) and bismuth (1.100 Å).27
The compositions of the compounds prepared
were analyzed by the ICP emission spectrometer.
As wastage and volatilization are unavoidable dur-
ing spark plasma sintering, the real composition
may deviate from the nominal one. However,
according to analysis results, content of various
elements really corresponds to the Bi2Te3,
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 compositions
(Table I).
Clear crystalline grained structures with mean
grain size of  1 lm were observed by the SEM
method for all the compositions. No remarkable
changes in the grain structures were found for
Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3. The SEM
images of the samples surfaces are shown in Fig. 2.
The densities of the bulk samples were equal to
6.72 g/cm3, 6.82 g/cm3, and 6.75 g/cm3 for Bi2Te3,
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3, respectively, which
are 87%, 88.5%, and 88% of the theoretical density
of Bi2Te3 (7.7 g/cm3). In fact, the densities of the
samples under study are less than the theoretical
density. However, it is important to note these
densities are weakly dependent on the composition
of compounds. The main purpose of this paper is to
find the of the Lu and Tm doping on the thermo-
electric properties of Bi2Te3. So, the densities of
both undoped and doped Bi2Te3 should be the same
to find this effect.
To confirm the uniformity of elements distribu-
tion, the EBSD method was applied. The results of
EBSD for Bi1.9Lu0.1Te3 are presented in Fig. 3. One
can see from the EBSD mapping that the chemical
elements are uniformly distributed.
So, both phase compositions and grain structures
of Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te are the
same. Then, sufficient changes of the thermoelectric
properties of these compounds discussed further
should be attributed to the Lu and Tm doping effect.
Fig. 2. SEM images on the Bi2Te3 (a), Bi1.9Lu0.1Te3 (b) and
According to the Hall effect study, the majority Bi1.9Tm0.1Te3 (c) surfaces.
charge carriers for all the compounds are electrons.
It is known28–30 that the type and carriers concen-
tration in Bi2Te3 are closely related to point defects.
The most common defects are antisite defects of Bi
at the Te sites (negatively charged Bi0Te , accompa- holes per defect). Since the energy of evaporation for
nied with formation of one hole, hÆ), vacancies at the Te (52.55 kJ/mol) is much lower than that of Bi
Te sites (positively charged VTe , providing two (104.80 kJ/mol), the evaporation of Te is much
electrons, e¢, per defect), and vacancies at the Bi easier than that of Bi. Each VTe vacancy leaves
sites (negatively charged V000 two free electrons, as described in Eq. 1
Bi , contributing three

Fig. 3. EBSD mapping of chemical elements on the Bi1.9Lu0.1Te3
surface.
Bi2 Te3 ¼ 2Bi   0
Bi þ 2TeTe þ TeðgÞ þ V Te þ 2e ; ð1Þ
where symbol g is corresponding to a gaseous phase.
The V000 
Bi and V Te vacancies always have a ratio
equal to 2:3, resulting in a zero net free charge, as
shown in Eq. 2

was mentioned above, ionic radii of the Lu3+ and
5Bi2 Te3 ¼ 8Bi  000 
Bi þ 10TeTe þ 5TeðgÞ þ 2VBi þ 3VTe
0  ð2Þ
þ 2BiTe þ 2h :
Due to small differences in electronegativity
between Te and Bi, antisite defects are induced
since Bi can easily jump from Bi site to Te site
contributing one hole as a free carrier (Eq. 2).
For polycrystalline samples, the dangling bonds
at grain boundaries due to Te deficiencies can be
considered as fractional-working as n-type dopants
in the same manner as the whole-defects inside the
grains. This is the reason why the most polycrys-
talline Bi2Te3 samples are n-type semiconductors.
Moreover, for the polycrystalline samples prepared
by the deformation methods including the ball
milling, hot pressing and spark plasma sintering,
the deformation can induce a donor-like effect to
take place. In these methods, the deformation can
result in the non-basal slip and produce, on average,
3 Te to 2 Bi vacancy-interstitial pairs.29 When
abundant Bi vacancies are created, the Bi atoms
occupying the Te sites would more readily move
back into its original sublattices and excess Te
vacancies are produced as described in Eq. 3
2V000   000   0
Bi þ 3V Te þ BiTe ¼ V Bi þ BiBi þ 4V Te þ 6e : ð3Þ
Yaprintsev, Lyubushkin, Soklakova, and Ivanov
Thus, six excess electrons are generated per
equation as an additional source of electrons.
Both vacancies (Eq. 1) and deformation resulting
in donor-like effect (Eq. 3) will generate electrons as
the majority charge carriers in Bi2Te3, Bi1.9Lu0.1Te3
and Bi1.9Tm0.1Te3.
The concentration and Hall mobility values of
electrons for these compounds taken at room tem-
perature are collected in Table II. One can see that
the Lu and Tm doping results in the increase of n
and decrease of lH. The doping effect on n is usually
attributed to the difference in electronegativity for
elements forming antisite defects responsible for
holes generation in Bi2Te3 (n accordance with
Eq. 2). The electronegativity values are equal to
2.1, 2.02, 1.27 and 1.25 for Te, Bi, Lu and Tm,
respectively. So, larger a electronegativity differ-
ence for the Lu–Te and Tm–Te pairs compared to
the Bi–Te pair will decrease the concentration of
antisite defects at the Te-sites, which contributes
one hole per defect and, hence, results in more
electrons. It is important to note the electronega-
tivity values are very close for Lu and Te. In this
case, the electron concentration for the Bi1.9Lu0.1Te3
and Bi1.9Tm0.1Te3 compounds should be really very
close (Table II).
Reducing the carrier mobility for Lu- or Tm-doped
Bi2Te3 can be related to an alloy scattering of
carriers.31,32 The alloy scattering the Bi1.9Lu0.1Te3
and Bi1.9Tm0.1Te3 compounds is related to forming
point defects in the Bi2Te3 lattice as a result of
substituting the Lu and Tm atoms for the Bi site. As
Tm3+ ions are very close. So, the lH difference for
the Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 compounds can-
not be attributed to various ionic radii of Lu3+ and
Tm3+. But, there is a remarkable difference in
magnetic properties of these ions. In fact, Tm3+ has
a magnetic moment equal to 7lB (lB is the Bohr
magneton), while Lu3+ has zero magnetic moment.
Hence, an additional electron scattering by the
magnetic moments of Tm3+ can result in appearance
of a new contribution to l in addition to the alloy
scattering. Then, the lH of the Tm-doped Bi2Te3
compound should be lower as compared to lH of the
Lu-doped compound.
The temperature dependences of the specific
electrical resistivity for Bi2Te3, Bi1.9Lu0.1Te3 and
Bi1.9Tm0.1Te3 are presented in Fig. 4a. As is seen, q
of all three samples increases with increasing
temperature. The specific electrical resistivity of
Bi2Te3 is effectively decreased by the Lu and Tm
doping within the temperature 290–630 K range.
This behavior is expected as the Lu and Tm atoms
behave as donors in the Bi2Te3 lattice.
To study the temperature q behavior in detail, the
dq/dT derivatives versus T dependences are plotted
in Fig. 4b. Clear maxima are observed in the dq/
dT(T) curves at temperature Td  470 K. These
Effects of Lu and Tm Doping on Thermoelectric Properties of Bi2Te3 Compound

Table II. Concentrations, Hall mobilities and density-of-states effective mass of majority carriers of Bi2Te3,
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3
Compound n, 1019 (cm23) lH (cm2 V21 s21) m

Bi2Te3 1.2 420 0.16m0

Bi1.9Tm0.1Te3 2.3 300 0.25m0
Bi1.9Lu0.1Te3 2.4 360 0.25m0

Fig. 4. The q versus T (a) and dq/dT versus T (b) for Bi2Te3 (curve
1), Bi1.9Lu0.1Te3 (2) and Bi1.9Tm0.1Te3 (3).

maxima can be related to the change of conductivity
mechanism. It is known that the specific electrical
resistivity of donor semiconductors is expressed17
1 Above room temperature, the dominant mechanism
q¼ ; ð4Þ
eln
where l is the electron mobility usually not equal to
the Hall mobility.
So, the temperature q changes in Fig. 4 could be
attributed to the l and/or n changes. The Bi2Te3
compound is known to be a degenerate semiconduc-
tor. The degenerate semiconductors are character-
ized by the T-independent concentration of carriers.
In this case, the temperature q behaviour will be
determined by the temperature l behaviour.
For our experiments, the q(T) behaviour at tem-
peratures below Td corresponds to a regime of the
degenerate semiconductor. There are several
Fig. 5. The q versus T2.2 (a) and Dq versus T (b) dependences for
Bi2Te3 (curve 1), Bi1.9Lu0.1Te3 (2) and Bi1.9Tm0.1Te3 (3).
mechanisms determining the l(T) dependence.
is acoustic and optical phonon scattering of carri-
ers.33 Acoustic phonon scattering acting as the main
scattering mechanism at low temperatures results
in the l  T3/2 dependence. This dependence is
valid until a contribution of optical phonon scatter-
ing can be neglected. Above the Debye temperature
(the Debye temperature is  155 K for Bi2Te3),
optical phonon scattering becomes comparable to
acoustic phonon scattering, and the temperature
dependence of the carrier mobility can be described
by an empirical expression given as
l  T m ; ð5Þ
with 1.5 < m < 2.5.

For instance, the electron mobility for n-type
silicon varies as T2.3 when both optical and acous-
tic phonon scattering become dominant.
According to Fig. 5a, below Td the best fit for the
experimental q(T) curves presented in Fig. 4a cor-
responds to expression (5) with m = 2.2.
For the high-temperature range above Td, the
q(T) curves start to deviate from the q(T)  T 2.2
law. This behaviour can be attributed to an onset of
intrinsic conductivity. In this case, a thermal exci-
tation of the charge carriers from valence band to
conduction band will generate both electrons in the
conduction band and holes in the valence band that
results in the q decrease in accordance with expres-
sion (4). To distinguish an intrinsic conductivity
contribution, Dq(T), the experimental q(T) curves
should be subtracted from the background q(T)  T
2.2
dependences shown as dashed lines in Fig. 5a.
The Dq(T) dependences above Td are presented in
Fig. 5b. These dependences are really the same for
all the compositions. So, no noticeable doping effect
on intrinsic conductivity of Bi2Te3 could be found in
our experiments. It can mean that a band gap does
not change at doping.
The temperature dependences of the Seebeck
coefficient for Bi2Te3, Bi1.9Lu0.1Te3 and
Bi1.9Tm0.1Te3 are shown in Fig. 6. Since the major-
ity charge carriers in these compounds are elec-
trons, the Seebeck coefficient has a negative sign.
Moreover, all the S(T) curves have a maximum.
Such maximum is typical for the Y-, Lu-, Ce-, Sm-
doped Bi2Te3 compounds.9,11,12,14,15 Usually, such a
kind of extreme S change is related to a bipolar
effect when two types of the charge carriers are
present due to intrinsic conductivity. As a rule,
thermal excitation of carriers induced by intrinsic
conductivity does not change too much the concen-
tration of the majority charge carriers, but increases
the minority carrier concentration. The Seebeck
coefficient of electron conductivity is negative,
whereas hole conductivity is characterized by the
positive Seebeck coefficient. Competition of these
two contributions with opposite S sign will form the
S(T) maximum in Fig. 6.
Fig. 6. The S versus T dependences for Bi2Te3, Bi1.9Lu0.1Te3 and
Bi1.9Tm0.1Te3.
Yaprintsev, Lyubushkin, Soklakova, and Ivanov
It is known17 that the Seebeck coefficient of the
degenerate semiconductors can be expressed as
 
2k2 Tm  p 2=3 3
S¼ B 2 þc ; ð6Þ
3eh 3n 2
where kB is the Boltzmann’s constant,  h is the
reduced Planck constant, m is the density-of-states
effective mass of electrons and c is the scattering
factor.
Expression (6) shows that a higher concentration
of electrons decreases the S value, while a larger
scattering factor increases the Seebeck coefficient.
Normally, the q increase will be accompanied by
increase of S. But Fig. 6 shows an opposite trend,
that is, the q increase results in the S decrease.
So, the possible m and c changes in addition to
the n change should be taken into account to explain
the S behavior of Bi2Te3 at the doping. The c value is
determined by the carriers scattering mechanism.
According to Fig. 5a, this mechanism is the same for
all the compositions and can be described by expres-
sion (5). So, c will be the same, too. The c value is
equal to  1/2 for acoustic phonon scattering and to
0 for optical phonon scattering above the Debye
temperature.9 As was discussed above, both optical
and acoustic phonon scattering should be consid-
ered as dominant mechanisms to explain the q(T)
behaviour of Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3
below Td. Therefore, for further S analysis, let us
assume that c = 1/2 (acoustic phonon scattering)
+0 (optical phonon scattering) = 1/2.
Next, in accordance with expression (6), S linearly
increases as T increases. As is shown by dashed
lines in Fig. 6, such kinds of linear T-dependences of
S are really observed for initial temperatures from
290 K up to  370 K. A rate of the linear S(T)
growth can be determined by a coefficient DS [lV/
K]/DT [K]  2.14 Æ 107. Using the n (Table II) DS/
DT and c values, the density-of-states effective mass
of electrons can be estimated. The m estimates are
given in Table II (m0 is the mass of the free
electron). So, m substantially increases at the
doping from 0.16m0 for undoped Bi2Te3 up to
0.25m0 for Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3. The m
increase can be related to forming a flat and narrow
impurity band with high density of states near the
Fermi level.34 This doping effect was successfully
used to explain improving ZT in Tl-doped PbTe.35
It is known that the 4f levels of electrons can form
the narrow and non-parabolic band lying near the
Fermi energy, increasing the density of states in the
compounds containing rare earth elements.3,36 Such
an impurity band could be believed to be one of the
sources resulting in the m and S increase in Lu-
and Tm doped Bi2Te3.
The temperature dependences of the power factor,
S2/q, for Bi2Te3, Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 are
presented in Fig. 7. These dependences combine the
q and S contributions. Although the q and S values
for Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 are quite
Effects of Lu and Tm Doping on Thermoelectric Properties of Bi2Te3 Compound
Fig. 7. The S 2/q versus T dependences for Bi2Te3, Bi1.9Lu0.1Te3
and Bi1.9Tm0.1Te3.
Fig. 8. (a) The k versus T dependences for Bi2Te3, Bi1.9Lu0.1Te3 and
Bi1.9Tm0.1Te3; (b) the klatt versus T and kcarr versus T dependences
for Bi2Te3.
different, the S2/q values of these compounds are
very close and much larger as compared to Bi2Te3.
The temperature dependences of the thermal
conductivity of all the samples are presented in
Fig. 8a. It was found that the Lu and Tm doping
decreases the thermal conductivity of Bi2Te3. The
Bi1.9Lu0.1Te3 compound has the lowest thermal
conductivity. In this case, the k value is close to
the thermal conductivity of the Y-, Ce and Sm-doped
Bi2Te3 compounds.19
The k behavior of the samples in Fig. 8a is in
contradiction to the q change, since the q decrease
usually results in the k increase. Such a q–k
relationship is usually stated by the Wiedemann–
Franz law, kcarr = LrT, where kcarr is the carrier
contribution to the thermal conductivity, L is the
Lorenz number and r = 1/q is the specific electrical
conductivity.37 Hence, the Wiedemann–Franz law is
a law only for the electron thermal conductivity. In
order to explain the above contradiction, the lattice
contribution to the thermal conductivity should be
taken into account.
It should be noted the Wiedemann–Franz law was
originally developed for metals and its use for
semiconductors can be limited. The Lorenz number
for metals is a constant equal to 2.45 9 108
W X K2. Let us apply the Wiedemann–Franz law
to determine both the electron, kcarr, and lattice,
klatt, thermal conductivities for the samples studied.
The temperature kcarr and klatt dependences for
Bi2Te3 are shown in Fig. 8b. The kcarr contribution
seems to be too large.38 So, the Lorenz number for
metals results in an incorrect kcarr estimate. It is
important too that the Wiedemann–Franz cannot
correctly distinguish the contributions from kcarr
and klatt in many semiconductors, in which the
Lorenz number depends on carrier density and
electron scattering.39,40 Moreover, the Wiedemann–
Franz calculation of the electron and lattice thermal
conductivities of Lu0.1Bi1.9Te3 and Lu0.1Bi1.9Te3
gives an unacceptable conclusion that klatt tends to
zero, if the Lorenz number equal to 2.45 9 108
W X K2 was assumed. So, the kcarr and klatt
contributions cannot be determined for these
compositions.
Several mechanisms reducing the thermal con-
ductivity of Bi2Te3 at the Lu and Tm doping could be
considered. First, this doping can introduce a num-
ber of various point defects in the Bi2Te3 lattice like
the antisite defects and Lu and Tm atoms substi-
tuting for the Bi sites. These defects can reduce kcarr
by scattering phonons due to either mass contrast or
local strains. For instance, theoretically, k of Bi2Te3
can be decreased down to 20% by the antisite
defects.9 Second, besides the Lu and Tm doping
effect on the lattice thermal conductivity, reducing
the electron thermal conductivity was theoretically
predicted for semiconductors with a narrow impu-
rity band having high and sharp density-of-states
near the Fermi level. Formation of such a band
originated from electronic 4f-levels of Lu or Tm,
which was before assumed to explain the S behavior
(Fig. 6). The physical reason for the kcarr decrease is
that the heat carried by an electron is proportional
to the difference between its energy and the Fermi
energy. So, materials with narrow density-of-states
(DE/2 less than several kBT, where DE is width of
band), which ‘‘cut off’’ the high energy end of the
Fermi distribution, have low kcarr and the Wiede-
mann–Franz law loses validity.34

Fig. 9. The ZT versus T dependences for Bi2Te3, Bi1.9Lu0.1Te3 and
Bi1.9Tm0.1Te3.
Now, let us discuss the temperature dependences
of the thermal conductivity presented in Fig. 8. A
minimum in the k(T) dependences can be related to
the change of the thermal conductivity mechanism.
The thermal conductivity of solids decreases as
k  T1 as temperature increases above the Debye
temperature. Such behavior is due to the phonon–
phonon scattering. Obviously, this mechanism can
be responsible for the k(T) behavior below the k
minimum. At higher temperatures corresponding to
a regime of intrinsic conductivity, k starts to
abruptly increase. This k increase can be originated
from the bipolar effect.9,41 In this case, the electron–
hole pairs are thermally excited at the hot-side of
the sample due to an intrinsic conductivity process.
Then, these pairs as neutral formations move to the
cold-side. Finally, the electron–hole pairs disappear
due to a recombination process. Energy of recombi-
nation per one electron–hole pair equal or greater
than the band gap will be emerging as a phonon.
The temperature dependences of ZT for Bi2Te3,
Bi1.9Lu0.1Te3 and Bi1.9Tm0.1Te3 are shown in Fig. 9.
One can see that Bi1.9Lu0.1Te3 has the highest ZT
values over the whole measured temperature range,
which can be attributed to the highest Seebeck
coefficient and lowest thermal conductivity and
specific electrical resistivity. The maximum ZT
value is equal to  0.9 for the temperature 450–
500 K range. The bipolar effect of decreasing S and
increasing k limits further ZT increase at higher
temperatures.
CONCLUSION
Thus, the Lu and Tm doping results in sufficient
increase of the thermoelectric figure-of-merit
from  0.4 for undoped Bi2Te3 up to  0.7 for
Bi1.9Tm0.1Te3 and  0.9 for Bi1.9Lu0.1Te3. Doping
effects enhancing the thermoelectric efficiency of
Bi2Te3 are (1) the decrease of the specific electrical
resistivity via the increase of the electron concen-
tration, since Lu and Tu behave as donors in the
Bi2Te3 lattice, (2) the increase of the Seebeck
coefficient via the increase of the density-of-states
Yaprintsev, Lyubushkin, Soklakova, and Ivanov
effective mass for conduction band, and (3) the
decrease of the total thermal conductivity via form-
ing the point defects like the Bi or Te vacancies and
antisite defects.
ACKNOWLEDGEMENTS
This work was also financially supported by the
Ministry of Education and Science of the Russian
Federation under Project Nos. 3.6586.2017/BY and
03.G25.31.0246.
CONFLICT OF INTEREST
The authors declare that they have no conflict of
interest.
REFERENCES
1. G. Mahan, B. Sales, and J. Sharp, Phys. Today 50, 42
(1997).
2. T.M. Tritt, Science 283, 804 (1999).
3. Y.C. Lan, A.J. Minnich, G. Chen, and Z.F. Ren, Adv. Funct.
Mater. 20, 357 (2010).
4. G.J. Snyder and E.S. Toberer, Nat. Mater. 7, 105 (2008).
5. H. Kitagawa, T. Nagamori, T. Tatsuta, T. Kitamura, Y.
Shinohara, and Y. Noda, Scr. Mater. 49, 309 (2003).
6. D.B. Hyun, T.S. Oh, J.S. Hwang, J.D. Shim, and N.V.
Kolomoets, Scr. Mater. 40, 49 (1998).
7. S. Miura, Y. Satob, K. Fukuda, K. Nishimura, and K. Ike-
da, Mater. Sci. Eng. A 277, 244 (2000).
8. O. Ivanov, O. Maradudina, and R. Lyubushkin, J. Alloys
Compd. 586, 679 (2014).
9. W. Liu, X. Yan, G. Chen, and Z. Ren, Nano Energy 1, 42
(2012).
10. Y. Li, J. Jiang, G. Xu, W. Li, L. Zhou, Y. Li, and P. Cui, J.
Alloys Compd. 480, 954 (2009).
11. S.S. Kim, S. Yamamoto, and T. Aizawa, J. Alloys Compd.
375, 107 (2004).
12. Y. Morisaki, H. Araki, H. Kitagawa, M. Orihashi, K.
Hasezaki, and K. Kimura, Mater. Trans. 46, 2518 (2005).
13. Q. Zhang, S. Che, W. Liu, K. Lukas, and X. Yan, Nano
Energy 1, 183 (2012).
14. X.K. Duan, K.G. Hu, D.H. Ma, W.N. Zhang, Y.Z. Jiang, and
S.C. Guo, Rare Met. 34, 770 (2015).
15. P. Srivastava and K. Singh, Mater. Let. 136, 337 (2014).
16. B. Jarivala, D. Shah, and N.M. Ravindra, J. Electron.
Mater. 44, 1509 (2015).
17. J. Yang, F. Wu, Z. Zhu, L. Yao, H. Song, and X. Hu, J.
Alloys Compd. 619, 401 (2015).
18. X.H. Ji, X.B. Zhao, Y.H. Zhang, B.H. Lu, and H.L. Ni, J.
Alloys Compd. 387, 282 (2005).
19. F. Wu, H. Song, J. Jia, and X. Hu, Prog. Nat. Sci. Mater.
Int. 23, 408 (2013).
20. F. Wu, W. Shi, and X. Hu, Electron. Mater. Lett. 11, 127
(2015).
21. X.H. Ji, X.B. Zhao, Y.H. Zhang, B.H. Lu, and H.L. Ni,
Mater. Lett. 59, 682 (2005).
22. F. Wu, H.Z. Song, J.F. Jia, F. Gao, Y.J. Zhang, and X. Hu,
Phys. Stat. Sol. A 210, 1183 (2013).
23. W.Y. Shi, F. Wu, K.L. Wang, J.J. Yang, H.Z. Song, and X.J.
Hu, Electron. Mater. 43, 3162 (2014).
24. X.B. Zhao, Y.H. Zhang, and X.H. Ji, Inorg. Chem. Com-
mun. 7, 386 (2004).
25. Y. Deng, X.S. Zhou, G.D. Wei, J. Liu, C.W. Nan, and S.J.
Zhao, J. Phys. Chem. Solids 63, 2119 (2002).
26. X.B. Zhao, X.H. Ji, Y.H. Zhang, and B.H. Lu, J. Alloys
Compd. 368, 349 (2004).
27. Y.Q. Jia, J. Sol. State Chem. 95, 184 (1991).
28. Y. Pan, T.R. Wei, C.F. Wu, and J.F. Li, J. Mater. Chem. C
3, 10583 (2015).
Effects of Lu and Tm Doping on Thermoelectric Properties of Bi2Te3 Compound
29. L. Hu, T. Zhu, X. Liu, and X. Zhao, Adv. Funct. Mater. 24,
5211 (2014).
30. J. Suh, K.M. Yu, D. Fu, X. Liu, F. Yang, J. Fan, D.J. Smith,
Y.H. Zhang, J.K. Furdyna, C. Dames, W. Walukiewicz, and
J. Wu, Adv. Mater. 27, 3681 (2015).
31. D.C. Look, D.K. Lorance, J.R. Sizelove, C.E. Stutz, K.R.
Evans, and D.W. Whitson, J. Appl. Phys. 71, 260 (1992).
32. H.S. Bennett, J. Appl. Phys. 80, 3844 (1996).
33. S.S. Li, Semiconductor Physical Electronics, ed. S.S. Li
(New York: Springer, 2006), pp. 211–245.
34. H.J. Goldsmid, J. Electron. Mater. 41, 2126 (2012).
35. J.P. Heremans, V. Jovovic, E.S. Toberer, A. Saramat, K.
Kurosaki, A. Charoenphakdee, S. Yamanaka, and G.J.
Snyder, Science 321, 554 (2008).
36. T.M. Tritt and M.A. Subramanian, MRS Bull. 31, 188
(2006).
37. K. Termentzidis, O. Pokropyvnyy, M. Woda, S. Xiong, Y.
Chumakov, P. Cortona, and S. Volz, J. Appl. Phys. 113,
013506 (2013).
38. D.K. Chauhan, W.M. Notling, P.F.P. Poudeu, and K.L.
Stokes, Mater. Sci. Semicond. Proc. 40, 453 (2015).
39. T.M. Tritt, Thermal Conductivity: Theory, Properties, and
Applications (New York: Plenum Publisher, 2004),
p. 285.
40. C.M. Bhandari and D.M. Rowe, J. Phys. D Appl. Phys. 18,
873 (1985).
41. S. Wang, J. Yang, T. Toll, J. Yang, W. Zhang, and X. Tang,
Sci. Rep. 5, 10136-1 (2015).